Análisis bayesiano: conceptos y fundamentos.
The Bayesian Analysis is introduced from a conceptual framework. The Bayesian school of thought is confronted to the classical frequentist school and the virtues and defects of each are discussed. It is shown how the concept of probability as degree of belief from plausible reasoning may be introduced, presented as a generalization of the aristotelian logic when reasoning under uncertain conditions. Cox's theorem, which demonstrates that laws governing plausible reasoning are equivalent to laws governing probability calculus, is discussed. The likelihood concept is analyzed and Bayes' theorem which, together with the concept of probability as degree of belief are the pillars upon which Bayesian Analysis is based, is demonstrated. The concept of interchangeability is introduced and its theoretical consequences are analyzed. The concept of diffuse or noninformative distribution is considered in depth. The Maximum Entropy Principle as a procedure oriented to obtain priors including all possible information before getting new data is thoroughly discussed. The Reference Analysis is introduced and developed in detail as a methodology to obtain a priori distributions as diffuse as possible and a posteriori distributions essentially expressing all information contained in data. The asymptotic convergence of the a posteriori distribution is analyzed. The concept of hierarchical Bayes' Analysis is sketched. The basic concepts of the Decision Theory in the Bayesian framework are introduced. The problem of Model Comparison is shown. The numerical calculus algorithms most frequently used for a Bayesian Analysis are briefly discussed. Ideas on the Bayesian Analysis in stock assessment are introduced and uncertainty sources to be considered in an analysis are listed.
| Main Author: | |
|---|---|
| Format: | Book/Monograph/Conference Proceedings biblioteca |
| Language: | Spanish / Castilian |
| Published: |
Instituto Nacional de Investigación y Desarrollo Pesquero (INIDEP)
2012
|
| Subjects: | Recursos pesqueros, Predicción, Teoría de la probabilidad, Análisis estadístico, Modelos estadísticos, Evaluación de efectivos, Fishery resources, Prediction, Probability theory, Statistical analysis, Statistical models, Stock assessment, |
| Online Access: | http://hdl.handle.net/1834/17024 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|